Your working tree is the workspace.
Search, inspect, edit, validate, and review without moving the task into a detached chat sandbox.
StemCode works inside your real repository: understand unfamiliar code, plan changes, edit files, run validation, review diffs, and automate pull-request feedback across terminal, desktop, editors, and CI.
curl -fsSL https://raw.githubusercontent.com/rizwan3d/StemCode/master/scripts/install.sh | bash
Extensions → search “StemCode”
VIEW VS CODE EXTENSION ↗
Extensions → Manage Extensions → search “StemCode”
VIEW VISUAL STUDIO EXTENSION ↗
Windows x64 · macOS x64/arm64 · Linux x64/arm64
DOWNLOAD DESKTOP ↗
dotnet add package StemCode
VIEW STEMCODE ON NUGET ↗
The NuGet package is the StemCode library. Install the CLI with the CLI installers above.
███████╗ ████████╗ ███████╗ ███╗ ███╗ ██████╗ ██████╗ ██████╗ ███████╗ ██╔════╝ ╚══██╔══╝ ██╔════╝ ████╗ ████║ ██╔════╝ ██╔═══██╗ ██╔══██╗ ██╔════╝ ███████╗ ██║ █████╗ ██╔████╔██║ ██║ ██║ ██║ ██║ ██║ █████╗ ╚════██║ ██║ ██╔══╝ ██║╚██╔╝██║ ██║ ██║ ██║ ██║ ██║ ██╔══╝ ███████║ ██║ ███████╗ ██║ ╚═╝ ██║ ╚██████╗ ╚██████╔╝ ██████╔╝ ███████╗ ╚══════╝ ╚═╝ ╚══════╝ ╚═╝ ╚═╝ ╚═════╝ ╚═════╝ ╚═════╝ ╚══════╝
The checkout limiter is keyed only by IP. I’ll inspect the middleware, switch it to endpoint metadata, then run the API tests.
• click to collapseexport function rateLimit(options: RateLimitOptions) {const key = request.ip;const bucket = `${key}:${options.windowMs}`;return limiter.consume(bucket);const route = request.routeOptions.url ?? "unknown";const scope = options.scope ?? route;const key = `${scope}:${request.ip}`;return limiter.consume(key);The limiter now separates buckets by route. I’ll run the focused tests before the full API suite.
• click to collapseSearch, inspect, edit, validate, and review without moving the task into a detached chat sandbox.
Permissions and profiles decide what can run automatically, what should ask first, and what stays blocked.
CLI, desktop, VS Code, Visual Studio, JetBrains, ACP-compatible editors, and CI review share the same local agent.
LSP-powered symbols, definitions, references, diagnostics, rename previews, and graph-aware local indexing improve context quality.
Keep long-lived project memory in .stemcode/memory and reusable prompts in .stemcode/commands.
Apply targeted edits, patches, insertions, and search/replace operations with undo and redo for tracked changes.
Use ! for local shell commands, !! for background terminals, and Tab completion for file and directory paths.
Spin up delegated tasks with independent contexts while the main conversation stays focused on the work that matters.
The agent can ask multiple-choice, multi-select, and free-form questions when a decision needs human input.
Voice dictation provides another input path without changing the repository-first StemCode workflow.
Automatic AI git commits can stay limited to files StemCode changed and avoid workspaces that already have staged changes.
Published release assets include SHA256 checksums and GitHub artifact attestations for build provenance.
Keep the agent close to the repository, the tools, and the human decisions that matter.
StemCode combines repository-aware search with built-in LSP-powered intelligence for symbols, definitions, references, diagnostics, and rename previews.
Use real shell tools and surgical tracked edits in the current working tree. StemCode can write files, apply patches, insert lines, search and replace, run checks, and keep changes undoable.
! and !!The same local agent experience can move from interactive implementation to editor work and automated code review.
Interactive work, one-shot prompts, piped input, session resume, and automation-friendly output.
stemcodeChat, model controls, profiles, activity output, permission prompts, and undo/redo.
WINDOWS · macOS · LINUXChat, selected context, file review, diff review, and applying suggestions without leaving VS Code.
EDITOR EXTENSIONUse the local CLI from Visual Studio through the Agent Client Protocol.
ACPA JetBrains tool window powered by the local CLI over ACP.
ACPRun StemCode in GitHub Actions, GitLab CI, and Bitbucket Pipelines for proposed changes.
PR · MR · PIPELINEInstall the published StemCode library from NuGet when you need reusable StemCode components in application code.
Control file edits, shell commands, network access, MCP tools, memory writes, and elevated operations.
Separate implementation, planning, review, exploration, and delegated work.
Store reusable commands and long-lived knowledge in versionable .stemcode/ files.
Use subscription-style sign-in, API-key providers, OpenAI-compatible endpoints, or local-model setups.
Search the repo, inspect files, follow symbols, and build focused context.
Turn a bug or feature request into concrete implementation steps.
Apply tracked edits, run tools, and iterate directly in the working tree.
Run the relevant shell commands, diagnostics, and test workflows.
Inspect files, diffs, pull requests, and merge requests with a findings-first workflow.
StemCode stays open source. Bring your own provider or local model for a $0 platform fee, or use optional StemCode Credits for managed model access. Credit usage varies by model and workload.
Run StemCode with your own API key, subscription sign-in, OpenAI-compatible endpoint, or local model.
A compact credit pack for focused fixes, reviews, and short implementation sessions.
More runway for everyday feature work, repository exploration, refactors, and validation loops.
For heavier coding workloads, long agent sessions, large repositories, and repeated review cycles.
The CLI, desktop releases, editor integrations, NuGet library, source, and documentation are all available from the public project.